Age | Commit message (Collapse) | Author | Files | Lines |
|
Would like to make a variant of the p10bmc machine that includes some of
the new etcd function. To do that, the search paths need to not include
the MACHINE variable. bitbake is smart enough to look under the
sub-directory and find the appropriate files.
Change-Id: I542b4568c18210870b3e621461585a4c7eee3121
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
|
|
Andrew Geissler (1):
systemd: no installation in templated targets
Includes needed service installation.
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: I22023bee9be60928a1ab990d627dc3d5553f30ad
|
|
Similar to previous generation of the POWER processor, introduce a
p10.inc that can be included on P10 systems.
Without this, the needed op-proc-control services and functions are not
brought in.
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: I58490e95e69ef8b4b31b0f4bc9213787058f0bba
|
|
Witherspoon uses UBI instead of static partitioning for flash
management, so make sure the qemu helper script can find the flash
image.
Tested: Reached a login prompt with `runqemu nographic slirp`.
Change-Id: I513d0a5b878c553456d70bef25adedfa25bf9c90
Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
|
|
Andrew Geissler (1):
systemd: no installation in templated targets
Includes changes required to ensure services are properly installed.
Tested:
-Confirmed services were installed into correct targets when building
p10bmc machine
Change-Id: I2d675ba083b2901ad1c26485d2f8b075fd6b3a25
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
This reverts commit ec7cab9378f548e94ee4af83946b2583d1fc7194.
ec7cab9378f5 ("Add socket-id for the first console") was only a partial
fixup of the obmc-console configurations found in the tree. For
historical reasons not all platforms that support multiple consoles
install a client configuration file for all consoles. Instead they relied
on some default behaviour that was removed in obmc-console's 4e7186918599
("Fixed broken dbus interface for multiple consoles"). Applying the
configuration changes in the manner of ec7cab9378f5 ("Add socket-id for
the first console") resulted in an asymmetric configuration between the
client and server, breaking some SSH SOL instances.
As of ae2460d0b8e8 ("obmc-console: Provide a default value for
`console-id`.") in obmc-console the requirement to specify
`socket-id`[^1] is lifted. Instead, the configuration can choose to
override a default value. This restores the original behaviour and
unbreaks SSH SOL.
[^1]: Now an alias of the `console-id` configuration key
Websocket and IPMI consoles remain broken. A fix is under development
that also takes a proper approach to supporting multiple consoles in
bmcweb, and the same solution pattern can be applied in ipmid.
More discussion of the problems involved can be found here:
https://amboar.github.io/notes/2023/05/08/happenings-in-obmc-console.html
Change-Id: I274284c791758f336da6d6301dc523bac2b5dd69
Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
|
|
Upstream yocto introduced a change via e510222 (systemd-systemctl:
fix instance template WantedBy symlink construction).
This fixes a bug that we in OpenBMC had been taking advantage of in that
we were able to document our templated target dependencies without it
actually doing anything. The real installation of services within
targets occurs in our bitbake recipes due to the complexity of chassis
and host instances on a per machine basis.
Leave the dependency information in the service files but comment them
out. It's useful to be able to look at a service and understand which
targets it's going to be installed into by the bitbake recipes.
Change-Id: I1bf3bc6ce7a79d6c5847f5f1b4e6a97ca67332f8
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
|
|
Upstream yocto introduced a change via e510222 (systemd-systemctl:
fix instance template WantedBy symlink construction).
This fixes a bug that we in OpenBMC had been taking advantage of in that
we were able to document our templated target dependencies without it
actually doing anything. The real installation of services within
targets occurs in our bitbake recipes due to the complexity of chassis
and host instances on a per machine basis.
Leave the dependency information in the service files but comment them
out. It's useful to be able to look at a service and understand which
targets it's going to be installed into by the bitbake recipes.
Change-Id: I9c485cbd19eb2804efe9916d4080ba3c662f65e5
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
|
|
Patrick Williams (1):
clang-format: copy latest and re-format
Change-Id: Ib3888ded06c8d0766b158293420b64644d600d7e
Signed-off-by: Andrew Geissler <openbmcbump-github@yahoo.com>
|
|
This corrects the associations between processor and OCC.
The OCC is power_managing the processor.
The processor is power_managed_by the OCC.
Retrieving the element that OCC is managing:
as 1 "/xyz/openbmc_project/inventory/system/chassis/motherboard/dcm0/cpu0"
Signed-off-by: Chris Cain <cjcain@us.ibm.com>
Change-Id: I45bce16127c34c70dadc133530ece0ed15c10b98
|
|
bmcweb now supports to specify the logging level by
https://gerrit.openbmc.org/c/openbmc/bmcweb/+/60212
This is to enable the Error or higher bmcweb logging levels
for meta-ibm.
Tested:
- Build BMC image
- Run bmcweb and verify Error/Critical logging messages
- Verify Redfish Validator
Change-Id: Ic431eafbaed22d25391201149976bd173a553c4e
Signed-off-by: Myung Bae <myungbae@us.ibm.com>
|
|
-mac address wasn't persisting after reboot
Tested by:
1. Assign a valid MAC address aa:e2:84:14:28:79 to the system using
curl -k -H "X-Auth-Token:${bmc_token}" -X PATCH -D patch.txt -d '{"MACAddress":"AA:E2:84:14:28:79"}' https://$bmc/redfish/v1/Managers/bmc/EthernetInterfaces/eth0
2. Now GET and verify if the MAC address has changed using
curl -k -H "X-Auth-Token:${bmc_token}" -X GET https://${bmc}/redfish/v1/Managers/bmc/EthernetInterfaces/eth0
3. Now reboot the BMC and verify if the MAC address remains the same
Change-Id: I334f1936f433e3e7e30ca5e669a89112f9879276
Signed-off-by: Isha Sahni <isha.sahni1@ibm.com>
|
|
Currently, scripts in install_ibm_plugins link_ibm_plugins are
not getting added in BMC image as a result of which changing "+="
to append.
Tested:
Post change can see the scripts are getting added in the image.
Change-Id: I932257e3652d0d6911923d4e9e0a57abeeb392ce
Signed-off-by: Chirag Sharma <chirshar@in.ibm.com>
|
|
This commit adds the rsyslogd process in P10BMC systems.
This is the revert of the commit -
https://gerrit.openbmc.org/c/openbmc/openbmc/+/59038
The OpenBMC community has defaulted to rsyslog
as the more feature rich alternative to syslog.
Even though we don’t need the remote offload capabilities of
rsyslog, it provides other features
(like forwarding other users of syslog to the journal).
Change-Id: I5f114d0157ec4d6c2390dccc5972019647f1ab20
Signed-off-by: Pavithra Barithaya <pavithra.b@ibm.com>
|
|
Adding an association between the OCCs and processors, so that the
throttle status can be associated with the applicable processor.
The processor is power_managed_by the OCC.
The OCC is power_managing the processor.
Change-Id: I4f5b9f4ec347665316dae87e1995cebb52ef374c
Signed-off-by: Chris Cain <cjcain@us.ibm.com>
|
|
This drop adds socket-id for first console. This was left out before
but with the new design this parameter is required and enforced by
obmc-console code. It is decided to set "socket-id = console0"
initially because that is the current path used by bmcweb but in future
you can change this value to whatever is appropriate.
The console information is now available through the DBUS interface.
Each console server register it's object path on DBUS. The leaf node of
the object path is used to create a console path. For example in the
following console0 the leaf node "console0" is used to make up the
console path and the GUI will use console path "/console/console0" to
connect to the console.
busctl tree xyz.openbmc_project.Console.console0
`-/xyz
`-/xyz/openbmc_project
`-/xyz/openbmc_project/console
`-/xyz/openbmc_project/console/console0
Here is a discord thread where new design was discussed.
https://discord.com/channels/775381525260664832/1083551792094249051
Tested:
Tested on the rainier system.
Related commits:
1) phosphor-dbus-interface: https://gerrit.openbmc.org/c/openbmc/phosphor-dbus-interfaces/+/61486
2) obmc-console: https://gerrit.openbmc.org/c/openbmc/obmc-console/+/62496
3) bmcweb: https://gerrit.openbmc.org/c/openbmc/bmcweb/+/62525
Change-Id: I5fd06f3aac69872a3e5032d592a9b59711795cfe
Signed-off-by: Ninad Palsule <ninadpalsule@us.ibm.com>
|
|
Upstream obmc-console now supports the meson build system. I consider
the autotools build system deprecated and intend to remove it.
Change-Id: Ic8eea195bee0b3bea1cb9418593596ce1aaceeae
Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
|
|
Commit is intended to enable import of trace
parser modules from hostboot repo. It aims
to pullin modules which parse trace section
of eLog committed by XGPE, PGPE and QME.
Change-Id: I067af7b9594df13dab7184e568d0d008d4e47c65
Signed-off-by: Prem Shanker Jha <premjha2@in.ibm.com>
|
|
Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: Iced76041c1207dc9ab28a83fd849ba3cd0a07bb5
|
|
Add the power-supply configuration files for the acbel-fsg032 power
supplies used in the Bonnell systems in order to get telemetry data
added to D-Bus via phosphor-hwmon.
Bonnell power supplies on bus 3 at 0x5a and 0x5b
Change-Id: I97a1e1ba57ee1b2da6462b6921f121683498fe87
Signed-off-by: Lakshmi Yadlapati <lakshmiy@us.ibm.com>
|
|
Enable Bonnell power supply driver acbel-fsg032
Change-Id: I271682b659565d5eb679ce624fc2fcc6b4b5c8c0
Signed-off-by: Lakshmi Yadlapati <lakshmiy@us.ibm.com>
|
|
In a base bb file where we are setting the primary git repository, it
is rare that we need to `SRC_URI +=`. This is an unnecessary pattern
that seems to have been copied throughout the repository. Remove the
pattern where appropriate and simply set SRC_URI directly.
Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: I430186a82f9582ba6196f5bf66b659af4092b48d
|
|
The witherspoon-tacoma server has the larger flash and will be utilized
for some OpenBMC prototype work. Add in some useful debug tools.
Change-Id: Ie3276088b1137a838a72e22a68a2c9fcf9bacb82
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
|
|
Tested:
- Confirmed tools are in p10bmc flash image
Change-Id: I808f38f410eaaad3077bb7eeaf36a166faf5beb4
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
|
|
The commit adds chassis association for cable cards. This change
is with respect to the downstream change where duplicate entry
for cable card association has been removed.
It was found that chassis association is missing upstream.
Signed-off-by: Sunny Srivastava <sunnsr25@in.ibm.com>
Change-Id: I2bb9c80aa50f29e67bba3abe9d5e8fa7de89f4d7
|
|
IBM is building a system under the "p10bmc" bitbake machine that only
has a single ethernet port, eth0, wired up on it. The system is still
utilizing an AST2600, but only wired eth0 out of the system for cost
purposes.
Since the "p10bmc" machine is a shared target there is no way to remove
services that assume eth1 is present during the build process. Instead,
rely on the fact that the device will be present in /sys/class/net/
if in fact the system supports the network device.
Tested:
- Verified that a p10bmc Rainier machine with 2 eth ports still ran this
script for both ports
- Verified that a p10bmc Bonnell machine with only eth0 ran this script
for eth0 but not eth1
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: Ic5dfe4bc772fb4de5eb28f3035cf9952d4bbcb7a
|
|
This is added in the line above as the witherspoon-tacoma is included in
the ibm-ac-server group. This package does not work on a tacoma though
(when building on POWER hardware) so remove it as it is not critical to
the function we need on a tacoma.
This does not build on POWER hardware due to this QEMU bug:
https://gitlab.com/qemu-project/qemu/-/issues/1528
Under the covers, the liberation-fonts repo uses fontcache.bbclass which
utilizes qemu-arm in the cross-compile environment.
Once the following bug is in a QEMU recipe, this will resolve the issue
but there's no reason to include this package on tacoma.
Tested:
- Confirmed witherspoon-tacoma machine builds on POWER hardware
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: I98220a51b838498b572bda07b28ed483e9316889
|
|
This adds support for the IBM Genesis3 board.
It's an evaluation platform based on Aspeed AST2500 SoC powering
a 4 socket Intel Cascade Lake server.
The main difference to regular platforms is the non-existence of a
CPLD/FPGA.
Everything on this board is directly connected to and controlled by the
BMC including power sequence for PCH/CPU.
Change-Id: Ia040dca83fbe17c52d9aa87cbf45d90452c031ff
Signed-off-by: Patrick Rudolph <patrick.rudolph@9elements.com>
Signed-off-by: Arthur Heymans <arthur.heymans@9elements.com>
Signed-off-by: Naresh Solanki <Naresh.Solanki@9elements.com>
|
|
Increasing the bmc dump size for ibm systems as the
current size 200KB is not sufficient and have seen
larger files i.e. journals are getting skipped.
Hence increasing the max size of a dump to 20MB.
Tested:
-After increasing the size journals are getting collected
in BMC dumps.
Change-Id: I0282bd4ae38831b80309caa59d1e804ca3d2ff1f
Signed-off-by: Chirag Sharma <chirshar@in.ibm.com>
|
|
The hardware watchdog config was disabled with commit [1] because the
fan-watchdog.bb in meta-ibm uses the /dev/watchdog already, which
according to the kernel documentation[2] it's the same device as
/dev/watchdog0.
```
[1]: https://gerrit.openbmc.org/c/openbmc/openbmc/+/60829
[2]: https://www.kernel.org/doc/Documentation/watchdog/watchdog-kernel-api.txt
```
Update the hardware watchdog config to use the currently unused
watchdog1 device to be able to recover from systemd hangs.
Verified that all IBM and OpenPower device trees contain a wdt2 device.
Tested:
- With the change, the BMC reboots after 2 min from injecting a systemd
error:
Mar 09 20:53:30 witherspoon systemd[1]: Caught <ILL> from PID 552.
Mar 09 20:53:30 witherspoon systemd-coredump[562]: Due to PID 1 having
crashed coredump collection will now be turned off.
Mar 09 20:54:25 witherspoon kernel: watchdog: watchdog1: watchdog did
not stop!
Mar 09 20:54:22 witherspoon systemd[1]: Freezing execution.
Mar 09 20:55:57 witherspoon systemd-journald[132]: Failed to send
WATCHDOG=1 notification message: Connection refused
client_loop: send disconnect: Broken pipe
- Without the change, the BMC just hangs, it stops pinging, and never
reboots:
Mar 09 21:07:23 witherspoon systemd[1]: Caught <ILL> from PID 433.
Mar 09 21:07:24 witherspoon systemd-coredump[687]: Due to PID 1 having
crashed coredump collection will now be turned off.
Mar 09 21:08:07 witherspoon systemd[1]: Freezing execution.
Mar 09 21:08:41 witherspoon systemd-journald[120]: Failed to send
WATCHDOG=1 notification message: Connection refused
Mar 09 21:10:11 witherspoon systemd-journald[120]: Failed to send
WATCHDOG=1 notification message: Transport endpoint is not connected
Mar 09 21:11:41 witherspoon systemd-journald[120]: Failed to send
WATCHDOG=1 notification message: Transport endpoint is not connected
Mar 09 21:13:12 witherspoon systemd-journald[120]: Failed to send
WATCHDOG=1 notification message: Transport endpoint is not connected
Change-Id: I7850d23805c1cb5c0b84cac4add28df16fe648f5
Signed-off-by: Adriana Kobylak <anoo@us.ibm.com>
|
|
Enable the minimum ship level format for p10bmc systems to match a
string that starts with some characters such as "ibm", followed by 2
numbers for the processor version such as "10", followed by the tag
number.
Change-Id: I4f3572db75b4fc1bb3c90c8c952d428d329b5d3a
Signed-off-by: Adriana Kobylak <anoo@us.ibm.com>
|
|
While the max31785 is at the same I2C bus and address as on Rainier, it
only has two fans so it needs a unique hwmon conf file, which it has
because the name is different in the device tree.
Signed-off-by: Matt Spinler <spinler@us.ibm.com>
Change-Id: Ia2c77e4149f02fe0a9d335cdd0551fb4ddd75108
|
|
Update the association configuration of fault identify according to
the latest association interface[1].
[1] https://github.com/openbmc/phosphor-dbus-interfaces/commit/e7c10bd5cdab6cd898e5cad0daab8c1128adc0c8
Signed-off-by: George Liu <liuxiwei@inspur.com>
Change-Id: I166c3dee16d8e709eb96982743fe30688b95012f
|
|
Update the association configuration of identify according to
the latest association interface[1].
[1] https://github.com/openbmc/phosphor-dbus-interfaces/commit/9eb460c6cfc2d7c2feb29c86f4beb31d4c3d9250
Signed-off-by: George Liu <liuxiwei@inspur.com>
Change-Id: I65ef596f3011720b968462f54d7a3b60e7c84002
|
|
This is added in the line above as the witherspoon-tacoma is included in
the ibm-ac-server group. This package does not work on a tacoma though
so remove it.
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: Ibf1bc79d6e33646bfff1fa4f560b0b3a40ceb270
|
|
The witherspoon-tacoma system is going to be utilized for some rack
controller prototyping. Add entity-manager in as it's required for
prototype work.
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: Iee11a5dd554ad9a9bad4dbfb44bdef7381a1101e
|
|
Set the BMC dump total allowed size as 409600KB.
p10bmc machines have an emmc based filesystem so therefore can fit more
bmc dumps.
Signed-off-by: Dhruvaraj Subhashchandran <dhruvaraj@in.ibm.com>
Change-Id: I05c1626ce9a03685adfdfd9713b3053ae8e63737
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
|
|
This feature requires special configuration and isn't used pervasively,
so turn it off by default.
Change-Id: Id7398df6c7511ceeb3a18a8759bd890ead3d1cc8
Signed-off-by: Brad Bishop <bradleyb@fuzziesquirrel.com>
|
|
Explicitly set FLASH_SIZE to prevent SPI-flash related configurations
being pulled into the build. Specifically this prevents the default
value provided by meta-phosphor triggering the uboot-flash-32768
override. The override pulls the u-boot_flash_32M.cfg configuration into
the build, breaking the environment handling for p10bmc machines which
exclusively use eMMC.
Change-Id: Ice0e2be27d5bd7ca207dcaf91ea311137fc55439
Signed-off-by: Andrew Jeffery <andrew@aj.id.au>
|
|
Change Ic46bcd9b8576530be7dcdda51384e1052d4a78f1 added hardware
watchdog support to all platforms but this conflicts with meta-ibm
support provided by fan-watchdog.bb. For the time being, supply a
disabled config file which will override the meta-phosphor version
so that systemd does not interact with the watchdog.
Change-Id: Ic6a587d8bab217cee8fbd421a8b6d6e3bd026631
|
|
Recent changes moved this binary from /usr/sbin to
/usr/libexec/phosphor-led-sysfs/. Update this service to reflect that.
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: I16df4cb1942ac93dfb8b487e58ff60ea67c58d28
|
|
As noted in the following mailing list thread:
https://lists.ozlabs.org/pipermail/openbmc/2022-November/032457.html
and implemented via this series of commits:
https://gerrit.openbmc.org/q/topic:op-power-control-ref
Some systemd services had their names changed. Update the service files
in this repo to reflect this.
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: Idaca9c1a46f8b5d673b4912e3027fc9268aa5dba
|
|
A recent refactoring requires the mmc include be first to ensure we
build an mmc based image
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: I7d4112dc5b39f0f21cc1dbc61e88b12ae3418215
|
|
IBM may have some prototyping use for our old witherspoon-tacoma systems
so get it building again.
clear_hostdumps_poweroff.service needs to be installed on tacoma
otherwise we get a "installed but not shipped" error.
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: I3ceeabe272636bc37bed040cc7e3b288c1fb15e5
|
|
Witherspoon requires UBI support in the kernel.
Signed-off-by: Joel Stanley <joel@jms.id.au>
Change-Id: Ifec95ca16bdf1792d68b2167e0d53f44e2f09239
|
|
Per association document[1], add/update the association between Fan
and Chassis.
[1] https://github.com/openbmc/phosphor-dbus-interfaces/commit/d0aa13b446ee1738d53fb1e04341946f5b0b7616
Signed-off-by: George Liu <liuxiwei@inspur.com>
Change-Id: I6c4d74b17fcfcbeb620832485928dcfd3460fc00
|
|
This is only used by certain IBM P10 and Tacoma systems. Let them
explicitly enable the driver.
Signed-off-by: Joel Stanley <joel@jms.id.au>
Change-Id: I384e5ab7587732f93617e05d06b700bc7421a81d
|
|
We have reached branch-capacity in the bb.utils.contains checks
at 3 image types. In order to be able to more easily add additional
image types, split the differences out into additional inc-snippets
for each image type.
Signed-off-by: Patrick Williams <patrick@stwcx.xyz>
Change-Id: Ibcdb5c885c67cb6d8728dfd315eb07f99b1a2173
|
|
IBM has run into some issues recently with certain network
configurations where eth1 is not getting set to 1 gigabit. This can
cause issues in scenarios where a network environment is running on 1
gigabit minimum-speed network switches.
Removing link-local support from eth1 seemed to aggravated this.
link-local is only on eth0 to avoid subnet issues. But without
link-local, the systemd call, systemd-networkd-wait-online, does not
always find an active eth1 so it never runs the command to force eth1 up
to 1 gigabit.
As far as we can tell, there's no appropriate mapping within
systemd-networkd-wait-online for it to know when the NCSI stack is
appropriately up on a network interface (allowing the 1 gigabit speed to
be set). Therefore the solution taken in this commit is to just keep
retrying for a period of time.
Tested:
- Verified new script ran and set eth ports as expected
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: I6737e208b59830262f51a8e30c78a97fc698e4fd
|
|
Signed-off-by: Andrew Geissler <geissonator@yahoo.com>
Change-Id: I8e54833ac78e540e9dd5011533d53ff9a3af6763
|